小波时变参数PGM(1,1)模型的工程应用研究

Application of Wavelet Time-Varying Parameter PGM(1,1)Model in the Prediction of Engineering

首先阐述了小波去除数据噪声的方法及小波模型的选取问题,然后对GM(1,1)模型背景值生成序列进行改进,采用搜索法确定背景值最佳生成系数。考虑到模型中的灰参数会随时间而变化,采用多项式逼近模型灰参数,并通过最小二乘方法求解多项式的系数,最后形成了小波时变参数PGM(1,1)模型同时采用Runge_Kutta方法求解。通过武咸城际铁路沉降监测数据的计算分析表明,小波时变参数PGM(1,1)在高铁沉降预测中具有较高的精度。

First,we elaborates the method of wavelet removing data noise and the selection of wavelet model. Then,improve the background value of GM（1,1）model,the optimal produce coefficient of which can be reached by searching method. Considering that the gray parameters in the model will change over time,we employ a polynomial to approach gray parameters of the model and the least square method to get the coefficients of polynomial. Finally,it develops into wavelet time-varying parameter PGM（1,1） model which is solved by Runge-Kutta method at the same time. Wu-Xian intercity railway subsidence monitoring data analysis shows that the wavelet time-varying parameter PGM（1,1）in highspeed rail subsidence has higher prediction accuracy.